Micro-Emitters Explained: Reliable Mid-Range Optics for AI Data Centers
Credo AVP of Product LK Bhupathi introduces ZeroFlap ALC with micro-emitter optics — a new take on optical interconnects. The wide-parallel approach spreads data across many small, low-power channels, building in redundancy, using far less power, and leaving plenty of headroom to scale for the next decade of AI infrastructure.
Built to Scale
Credo’s ZeroFlap ALCs extend reach up to 50 meters with dramatically reduced bulk—without sacrificing copper-class reliability. Powered by micro-emitter technology, Credo’s roadmap will support near-package and co-packaged connectivity solutions delivering ultra-dense bandwidth, exceptional power efficiency, and resilient future-ready connectivity.
Micro-Emitters Built for the Next Decade of AI Infrastructure
Scale-up and scale-out networks are getting faster and denser while demanding greater reliability and power efficiency. Scale-in interconnects are emerging as a new category of ultra-high throughput, chip-to-chip connections.
Scale-out networks
Scale-out networks connect dense GPU clusters and enable pipeline parallelism. Credo’s ALCs extend reach while reducing bulk and power without sacrificing reliability.
Scale-up networks
Scale-up networks are ~10x denser than scale-out and currently rely on passive copper within the rack. As they extend to row scale (10-30m), they will require ultra-reliable and low-power interconnects, making micro-emitter-based solutions ideal candidates.
Scale-in interconnects
Scale-in interconnects require ultra-high throughput and are ~10x denser than scale-up. These chip-to-chip interconnects between GPUs or GPUs and memory may need only up to 1m reach. Micro-emitter-based interconnects can scale down in size and deliver the densities needed for this emerging application.